Praveen latest patch updates

AI-Travel-Agent/requirements.txt

@@ -2,10 +2,10 @@ amadeus==12.0.0
 ipykernel==7.0.1
 jupyter==1.1.1
 langchain==1.0.2
-langchain-community==0.3.27
+langchain-community==0.4.1
 streamlit==1.51.0
 huggingface-hub==1.1.4
 pydantic==2.12.3
 wikipedia==1.4.0
 google-search-results==2.4.2
-hf-xet==1.2.0
+hf-xet==1.2.0

LLM/rag/requirements.txt

@@ -1,12 +1,12 @@
-langchain==0.3.24
+langchain==0.3.27
 langchain-community==0.3.27
-langchain-core==0.3.56
+langchain-core==0.3.72
 langchain-huggingface==0.1.2
-huggingface-hub==0.29.1
+huggingface-hub>=0.30.0
 sentence-transformers==3.4.1
 chromadb==0.6.3
 transformers==4.53.1
 pypdf==6.1.3
 torch==2.7.1
 langchain-chroma==0.2.2
-beautifulsoup4==4.13.3
+beautifulsoup4==4.13.3

LLM/src/README.md

@@ -0,0 +1,150 @@
+# RAG Chat with Ollama
+
+A Streamlit-based Retrieval-Augmented Generation (RAG) chat application powered by Ollama on Intel® Core™ Ultra Processors.
+
+## Overview
+
+This application demonstrates a RAG (Retrieval-Augmented Generation) system that allows you to chat with documents using Ollama's language models. Upload your documents, and the system will create embeddings and enable semantic search to provide context-aware responses.
+
+## Prerequisites
+
+- Windows 11 or Ubuntu 20.04+
+- Intel® Core™ Ultra Processors or Intel Arc™ Graphics
+- 16GB+ RAM recommended
+
+## Setup
+
+### 1. Install Ollama
+
+Download and install Ollama:
+```powershell
+winget install Ollama.Ollama
+```
+
+Or download from [https://ollama.com/download](https://ollama.com/download)
+
+### 2. Building Ollama with GPU Support (Vulkan)
+
+For advanced users who want to build Ollama from source with Vulkan GPU acceleration on Windows:
+
+**a. Install Vulkan SDK**
+- Download from: [https://vulkan.lunarg.com/sdk/home](https://vulkan.lunarg.com/sdk/home)
+
+**b. Install TDM-GCC**
+- Download from: [https://github.com/jmeubank/tdm-gcc/releases/tag/v10.3.0-tdm64-2](https://github.com/jmeubank/tdm-gcc/releases/tag/v10.3.0-tdm64-2)
+
+**c. Install Go SDK**
+- Download Go v1.24.9: [https://go.dev/dl/go1.24.9.windows-amd64.msi](https://go.dev/dl/go1.24.9.windows-amd64.msi)
+
+**d. Build Ollama with Vulkan**
+```powershell
+# Set environment variables
+set CGO_ENABLED=1
+set CGO_CFLAGS=-IC:\VulkanSDK\1.4.321.1\Include
+
+# Build with CMake
+cmake -B build
+cmake --build build --config Release -j14
+
+# Build Go binary
+go build
+
+# Run Ollama server (Terminal 1)
+go run . serve
+
+# Test with a model (Terminal 2)
+ollama run gemma3:270m
+```
+
+**Note:** This is for advanced users. The pre-built Ollama installation works fine for most users.
+
+### 3. Pull Language Models
+
+Pull the models you want to use:
+```bash
+ollama pull llama3.2
+ollama pull qwen2.5
+ollama pull mistral
+```
+
+### 4. Install Python Dependencies
+
+Using pip:
+```bash
+pip install streamlit ollama chromadb sentence-transformers pypdf
+```
+
+Using uv (recommended):
+```bash
+uv pip install streamlit ollama chromadb sentence-transformers pypdf
+```
+
+## Running the Application
+
+### 1. Start Ollama Server
+
+If not already running:
+```bash
+ollama serve
+```
+
+### 2. Run the Streamlit App
+
+```bash
+# Using Python directly
+streamlit run st_rag_chat.py
+
+# Or using uv
+uv run streamlit run st_rag_chat.py
+```
+
+### 3. Access the App
+
+Open your browser and navigate to:
+```
+http://localhost:8501
+```
+
+## Usage
+
+1. **Upload Documents**: Use the sidebar to upload PDF or text files
+2. **Select Model**: Choose your preferred Ollama model from the dropdown
+3. **Process Documents**: Click "Process Documents" to create embeddings
+4. **Chat**: Ask questions about your documents in the chat interface
+5. **View Sources**: See which document sections were used to answer your questions
+
+## Features
+
+- 📄 **Multi-format Support**: Upload PDF and text documents
+- 🤖 **Model Selection**: Choose from available Ollama models
+- 🔍 **Semantic Search**: Find relevant context using vector embeddings
+- 💬 **Context-Aware Chat**: Get answers based on your documents
+- 📚 **Source Attribution**: See which parts of documents were used
+- 💾 **Persistent Storage**: ChromaDB vector database for efficient retrieval
+
+## Troubleshooting
+
+**Ollama Connection Error:**
+- Ensure Ollama is running: `ollama serve`
+- Check if models are installed: `ollama list`
+
+**Memory Issues:**
+- Use smaller models like `llama3.2:1b` or `qwen2.5:3b`
+- Reduce the number of documents processed at once
+
+**Slow Performance:**
+- Ensure GPU drivers are up to date
+- Use GPU-accelerated Ollama build (Vulkan)
+- Try smaller, faster models
+
+## Technical Stack
+
+- **Ollama**: Local LLM runtime
+- **Streamlit**: Web interface
+- **ChromaDB**: Vector database
+- **Sentence Transformers**: Text embeddings
+- **Intel Hardware**: Optimized for Intel Core™ Ultra Processors
+
+## License
+
+This project is licensed under the MIT License. See [LICENSE](../LICENSE) for details.

LLM/src/st_ollama.py

@@ -13,7 +13,8 @@ def load_models():
         list: A list of model names if successful, otherwise an empty list.
     """
     try:
-        model_list = [model["name"] for model in ollama.list()["models"]]
+        response = ollama.list()
+        model_list = [model.model for model in response.models]
         return model_list
     except Exception as e:
         st.error(f"Error loading models: {e}")